blendify package

Subpackages

Submodules

blendify.scene module

class blendify.scene.Scene(*args, **kwargs)[source]

Bases: object

attach_blend(path: str | Path, with_camera: bool = False)[source]

Append objects and materials from the existing .blend file to the scene. The only two modalities that can be added to blendify Scene are lights and optionally camera, others (renderable objects, materials, etc.) are only appended. The appended modalities will only be present in the internal Blender structures, but will not be added to the blendify Scene class structure. However, they will appear on rendering and in the exported .blend files

Parameters:

  • path – path to the .blend file to append the contents from

  • with_camera – parse camera parameters from the .blend file or keep existing camera

property camera: Camera
static check_any_exists(fileprefix: str, filesuffixes: Sequence[str]) → bool[source]

Check if any of the combinations of <fileprefix>+<any filesuffix> exist in the filesystem

Parameters:

  • fileprefix (str) – single file prefix, can the full path or local name

  • filesuffixes (Sequence[str]) – a sequence of file suffixes to choose from

Returns:

True is any of the combinations exists in the filesystem, False otherwise

Return type:

bool

clear()[source]

Clears the scene

static export(path: str | Path, include_file_textures: bool = True, verbose: bool = False)[source]

Export the current scene to the .blend file

Parameters:

  • path (Union[str, Path]) – path to the target .blend file

  • include_file_textures (bool) – whether to write textures loaded from external files inside .blend file

  • verbose (bool) – whether to allow blender to log its status to stdout during exporting

preview(filepath: str | Path | None = None, save_depth: bool = False, save_albedo: bool = False, verbose: bool = False, fast: bool = False)[source]

Renders a scene using Blender’s OpenGL renderer. Linux and MacOS Only.

Parameters:

  • filepath (Union[str, Path]) – path to the image (PNG) to render to, returns the image as numpy array if None

  • use_gpu (bool) – whether to render on GPU or not

  • samples (bool) – number of raytracing samples per pixel

  • save_depth (bool) – whether to save the depth in the separate file. If yes, the numpy array <filepath>.depth.npy will be created if filepath is set, otherwise appends the array to the output.

  • save_albedo (bool) – whether to save albedo (raw color information) in the separate file. If yes, the PNG image <filepath>.albedo.png with color information will be created if filepath is set, otherwise appends the array to the output.

  • verbose (bool) – whether to allow blender to log its status to stdout during rendering

  • use_denoiser (bool) – use openimage denoiser to denoise the result

  • fast (bool) – whether to use fast and colorless preview mode (workbench engine)

static read_exr_distmap(path: str, dist_thresh: float = 10000.0) → ndarray[source]

Reads the distance map stored in EXR format, filters out all the values after a certain distance threshold. Requires OpenEXR to be installed in the system

Parameters:

  • path (str) – path to the .exr file

  • dist_thresh (float) – distance clip threshold

Returns:

distance map in numpy array format

Return type:

np.ndarray

static read_image(path: str) → ndarray[source]

Reads the image stored in PNG or JPG format

Parameters:

path (str) – path to the image file

Returns:

image in numpy array format

Return type:

np.ndarray

render(filepath: str | Path | None = None, use_gpu: bool = True, samples: int = 128, save_depth: bool = False, save_albedo: bool = False, verbose: bool = False, use_denoiser: bool = False, aa_filter_width: float = 1.5)[source]

Start the Blender rendering process

Automatically detects if shadow catcher objects are present in the scene by checking is_shadow_catcher property of Blender objects. For rendering with shadow catchers background is made white, because since Blender 3.0 shadow catcher rendering pass is made to be multiplied with the background.

Parameters:

  • filepath (Union[str, Path]) – path to the image (PNG) to render to, returns the image as numpy array if None

  • use_gpu (bool) – whether to render on GPU or not

  • samples (bool) – number of raytracing samples per pixel

  • save_depth (bool) – whether to save the depth in the separate file. If yes, the numpy array <filepath>.depth.npy will be created if filepath is set, otherwise appends the array to the output.

  • save_albedo (bool) – whether to save albedo (raw color information) in the separate file. If yes, the PNG image <filepath>.albedo.png with color information will be created if filepath is set, otherwise appends the array to the output.

  • verbose (bool) – whether to allow blender to log its status to stdout during rendering

  • use_denoiser (bool) – use openimage denoiser to denoise the result

  • aa_filter_width (float) – width of the anti-aliasing filter, set 0 to turn off

set_orthographic_camera(resolution: ndarray | Tuple[int, int], ortho_scale: float = 1.0, near: float = 0.1, far: float = 100.0, tag: str = 'camera', rotation_mode: str = 'quaternionWXYZ', rotation: ndarray | Tuple[float, float, float, float] | Tuple[float, float, float] | Tuple[Tuple[float, float, float], Tuple[float, float, float], Tuple[float, float, float]] | Positionable | Tuple[ndarray | Tuple[float, float, float], float] | Tuple[Positionable, float] = None, translation: ndarray | Tuple[float, float, float] = (0, 0, 0), resolution_percentage: int = 100) → OrthographicCamera[source]

Set orthographic camera in the scene. Replaces the previous scene camera, if it exists

Parameters:

  • resolution (Vector2di) – (w, h), the resolution of the resulting image

  • ortho_scale (float, optional) – Orthographic Camera scale (similar to zoom) (default: 1.0)

  • near (float, optional) – Camera near clipping distance (default: 0.1)

  • far (float, optional) – Camera far clipping distance (default: 100)

  • tag (str) – name of the created object in Blender

  • rotation_mode (str) – type of rotation representation. Can be one of the following: - “quaternionWXYZ” - WXYZ quaternion - “quaternionXYZW” - XYZW quaternion - “rotvec” - axis-angle representation of rotation - “rotmat” - 3x3 rotation matrix - “euler<mode>” - Euler angles with the specified order of rotation, e.g. XYZ, xyz, ZXZ, etc. Refer to scipy.spatial.transform.Rotation.from_euler for details. - “look_at” - look at rotation, the rotation is defined by the point to look at and, optional, the rotation around the forward direction vector (a single float value in tuple or list)

  • rotation (RotationParams) – rotation parameters according to the rotation_mode - for “quaternionWXYZ” and “quaternionXYZW” - Vec4d - for “rotvec” - Vec3d - for “rotmat” - Mat3x3 - for “euler<mode>” - Vec3d - for “look_at” - Vec3d, Positionable or Tuple[Vec3d/Positionable, float], where float is the rotation around the forward direction vector in degrees

  • translation (Vector3d, optional) – translation applied to the Blender object (default: (0,0,0))

  • resolution_percentage (int, optional)

Returns:

created camera

Return type:

OrthographicCamera

set_perspective_camera(resolution: ndarray | Tuple[int, int], focal_dist: float = None, fov_x: float = None, fov_y: float = None, center: ndarray | Tuple[float, float] = None, near: float = 0.1, far: float = 100.0, tag: str = 'camera', rotation_mode: str = 'quaternionWXYZ', rotation: ndarray | Tuple[float, float, float, float] | Tuple[float, float, float] | Tuple[Tuple[float, float, float], Tuple[float, float, float], Tuple[float, float, float]] | Positionable | Tuple[ndarray | Tuple[float, float, float], float] | Tuple[Positionable, float] = None, translation: ndarray | Tuple[float, float, float] = (0, 0, 0), resolution_percentage: int = 100) → PerspectiveCamera[source]

Set perspective camera in the scene. Replaces the previous scene camera, if it exists. One of focal_dist, fov_x or fov_y is required to set the camera parameters

Parameters:

  • resolution (Vector2di) – (w, h), the resolution of the resulting image

  • focal_dist (float, optional) – Perspective Camera focal distance in millimeters (default: None)

  • fov_x (float, optional) – Camera lens horizontal field of view (default: None)

  • fov_y (float, optional) – Camera lens vertical field of view (default: None)

  • center (Vector2d, optional) – (x, y), horizontal and vertical shifts of the Camera (default: None)

  • near (float, optional) – Camera near clipping distance (default: 0.1)

  • far (float, optional) – Camera far clipping distance (default: 100)

  • tag (str) – name of the created object in Blender

  • rotation_mode (str) – type of rotation representation. Can be one of the following: - “quaternionWXYZ” - WXYZ quaternion - “quaternionXYZW” - XYZW quaternion - “rotvec” - axis-angle representation of rotation - “rotmat” - 3x3 rotation matrix - “euler<mode>” - Euler angles with the specified order of rotation, e.g. XYZ, xyz, ZXZ, etc. Refer to scipy.spatial.transform.Rotation.from_euler for details. - “look_at” - look at rotation, the rotation is defined by the point to look at and, optional, the rotation around the forward direction vector (a single float value in tuple or list)

  • rotation (RotationParams) – rotation parameters according to the rotation_mode - for “quaternionWXYZ” and “quaternionXYZW” - Vec4d - for “rotvec” - Vec3d - for “rotmat” - Mat3x3 - for “euler<mode>” - Vec3d - for “look_at” - Vec3d, Positionable or Tuple[Vec3d/Positionable, float], where float is the rotation around the forward direction vector in degrees

  • translation (Vector3d, optional) – translation applied to the Blender object (default: (0,0,0))

  • resolution_percentage (int, optional)

Returns:

created camera

Return type:

PerspectiveCamera

Module contents